Designing Role Vectors to Improve LLM Inference Behaviour

TL;DR

This paper introduces role vectors derived from model activations as a novel method to steer LLM behavior, demonstrating their effectiveness in improving domain-specific performance over traditional persona prompts.

Contribution

The study presents a new approach using role vectors to influence LLM behavior, showing they outperform persona-based prompting in guiding models toward domain expertise.

Findings

Role vectors influence model behavior and improve task performance.

Activation addition reinforces role-specific directions.

Directional ablation removes influence, affecting performance.

Abstract

The influence of personas on Large Language Models (LLMs) has been widely studied, yet their direct impact on performance remains uncertain. This work explores a novel approach to guiding LLM behaviour through role vectors, an alternative to persona-based prompting. We construct 29 role vectors derived from model activations and evaluate their impact on benchmark performance across multiple domains. Our analysis investigates whether these vectors can effectively steer models toward domain-specific expertise. We measure two key interventions: (i) activation addition, which reinforces role-specific directions, and (ii) directional ablation, which removes them. Results on well-established benchmarks indicate that role vectors do, in fact, influence model behaviour, improving task performance in relevant domains while marginally affecting unrelated tasks. This, in turn, suggests that manipulating internal model representations has a greater impact on outcomes than persona-based prompting.